[
https://issues.apache.org/jira/browse/SOLR-10822?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16055709#comment-16055709
]
Noble Paul commented on SOLR-10822:
-----------------------------------
synchronizing just the computing part is not really useful. Overseer must
either wait till the collection admin operations are complete or there should
be mechanism to borrow session from an ongoing operation. We should warn users
in the policy framework documenation that running multiple ops in parallel may
lead to incorrect placements of replicas
> Concurrent replica placement calculations using Autoscaling Policy should not
> be allowed
> ----------------------------------------------------------------------------------------
>
> Key: SOLR-10822
> URL: https://issues.apache.org/jira/browse/SOLR-10822
> Project: Solr
> Issue Type: Sub-task
> Security Level: Public(Default Security Level. Issues are Public)
> Components: SolrCloud
> Reporter: Shalin Shekhar Mangar
> Labels: autoscaling
> Fix For: master (7.0)
>
>
> Policy framework are now used to find replica placements by all collection
> APIs but since these APIs can be executed concurrently, we can get wrong
> placements because of concurrently running calculations. We should
> synchronize just the calculation part so that they happen serially.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: dev-h...@lucene.apache.org
